Hi, we are using TAS5630DKD in a new product, and experiencing malfunction in about 7% of units. Description as follows:
- 3x TAS5630 per unit for 6x channels. One chip is the clock master and the other two slaved from it
- BTL configuration, overall circuit, configuration and layout very similar to the EVM board
- 39V PVDD; 12.2V GVDD
- Output filter inductor 100% identical to EVM using same core and construction
- 33nF X7R 25V 0603 bootstrap caps
In general, everything works fine and has even passed avionic qualification and accelerated life testing. However, on a total of 7 units out of the first production batch of 100, the OUT_B of always the same IC (one of the slaves) will not work, which is cause for concern for higher volume production.
Its OUT_C and OUT_D still works fine, but OUT_A and OUT_B are dead (0V). Their bootstap voltages seem fine at the correct DC offset. After deasserting /RESET, OUT_A begins switching for a few hundred ms, then output falls to zero. /SD is immediately asserted after /RESET is removed, RDY never goes high.
OUT_B does the same as OUT_A, but the initial switching waveform appears distorted. This is under ideal situations, with power rails already stabilized upon removing RESET, no output load connected and no input signal applied.
Since the other two bridges in the same IC keeps on working perfectly, most other possible issues are eliminated
Of all the units exhibiting this behaviour, it has always been on the same IC of the three, and always on the OUT_A and OUT_B. Replacing bootstrap caps did not help, and remarkably replacing the IC does not help either. This would point towards a PCB error, but all nodes on blank & erroneous PCB's have been tested with no short-circuits etc. found, and the layout for each IC is essentially a copy of the other two with no difference w.r.t. clearances, trace lengths, widths etc. Bootstrap caps are located right underneath the IC as on the EVM with 15mil vias.
It would therefore seem that something is driving at least the OUT_B half-bridge into shutdown, but I have no idea what it could be.
Any input would be greatly appreciated.
